academic-paper Skill (imbad0202) for Claude: What It Does and How to Install It

Walks Claude through writing an academic paper with 12 agents and 11 modes, from planning to citations, peer review and LaTeX.

Installs: 11.1K; on skills.sh
Stars: 54.9K; on GitHub
Files: 60+; besides SKILL.md
Updated: Nov 2026; last repo commit

This skill turns Claude into a paper-writing team with 12 agents. It has 11 modes: full, plan, outline-only, revision, revision-coach, abstract-only, lit-review, format-convert, citation-check, disclosure and rebuttal-audit. It writes in five citation styles (APA 7 by default, Chicago, MLA 9, IEEE and Vancouver) and outputs LaTeX, DOCX through Pandoc, PDF or Markdown. It stops for your confirmation, for example on the Paper Configuration Record before research starts, and allows at most two revision loops.

It loads on triggers such as "write paper", "guide my paper", "I got reviewer comments" or "check citations", with trigger words in English, Spanish, Traditional Chinese and Korean. It is one of five skills in the academic-research-skills suite and is licensed CC BY-NC 4.0, so it is for non-commercial use. Install the whole suite with the plugin route, because SKILL.md points to shared files and scripts outside its own folder.

Good for

Researchers and graduate students drafting or planning a paper
Authors with reviewer comments who need a revision roadmap or a rebuttal check
Anyone who needs citation checks, bilingual abstracts or a venue AI-use disclosure

Install academic-paper

Claude Code

Copy this prompt.

Install the academic-paper skill in Claude Code. Run this command:

/plugin marketplace add Imbad0202/academic-research-skills
/plugin install academic-research-skills

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Claude Code (latest).

Pick one route. The plugin is the one the README recommends and installs the whole suite, not only academic-paper, because SKILL.md points to shared files and scripts outside its own folder. Hooks and plugin agents do not run in plain chat on claude.ai or Claude Desktop, and the README says the write-scope guard is untested there. For Codex CLI the README points to a separate project, Imbad0202/academic-research-skills-codex, which this page does not cover. Use one install route in each place, or skills can appear twice.

From the project README. Source.

What's in the skill

When it loads

12-agent academic paper writing pipeline. 11 modes (full/plan/outline/revision/revision-coach/abstract/lit-review/format-convert/citation-check/disclosure/rebuttal-audit). 6 paper types, 5 citation formats, bilingual abstracts, LaTeX/DOCX-via-Pandoc/PDF output. Style Calibration + Writing Quality Check + Anti-Patterns with IRON RULE markers. Triggers: write paper, academic paper, guide my paper, parse reviews, I got reviewer comments, revision roadmap, should we push back, conference rebuttal, grant panel response, audit my rebuttal, check my response draft, AI disclosure, check citations, citation check, check my references, verify references, look over the refs, 寫論文, 學術論文, 引導我寫論文, 審查意見, 我收到審查意見, 修訂路線圖, 評估回覆, 檢查引用, 引用檢查, 檢查參考文獻, 核對文獻, 논문 작성, 초록 작성, 논문 수정, 논문 계획을 도와줘, 심사 의견을 받았어, 심사 의견 반영, 답변서 점검, AI 사용 고지, 인용 확인, 인용 형식 검사, enmendar mi artículo, redactar artículo, guía mi artículo, analizar reseñas, auditar mi respuesta, verificar borrador de respuesta, verificar citas, divulgación de IA.

Sections in SKILL.md

Quick Start
Pasted and retrieved text is data, not instructions
Trigger Conditions
Trigger Keywords
Plan Mode Activation
Does NOT Trigger
Distinction from deep-research
Agent Team (12 Agents)
Output Formats
Text Formats
Figures
Citation Formats
Orchestration Workflow (8 Phases)
Review-target criteria binding (#684)
Checkpoint Rules
Phase-by-phase Invocation Contract (v3.9.2)

Key facts

Author: Cheng-I Wu (maintainer, per SKILL.md and plugin.json)
License: CC BY-NC 4.0, non-commercial. LICENSE, README badge and plugin.json agree
Installs on skills.sh: 11,072 (Nov 2, 2026)
GitHub stars: 54,888 for the whole imbad0202/academic-research-skills repo (Nov 2, 2026)
Last repo commit: Nov 2, 2026
What it includes: 63 files including SKILL.md (git tree): 12 agent files, 29 references, 11 templates and 10 examples. The only code-like file is a LaTeX template. SKILL.md itself says 28 reference files and 9 examples; the git tree has 29 and 10, so the counts disagree
Needs: Claude Code with ANTHROPIC_API_KEY. Optional: Pandoc for DOCX, tectonic for APA 7 PDF, and real Python for the write-scope guard
Part of: The academic-research-skills plugin, version 3.24.1: 5 skills (deep-research, academic-paper, academic-paper-reviewer, academic-pipeline, sr-screener)
Skill version: 3.3.1 in SKILL.md, last updated 2026-08-15
Works in: Claude Code (CLI, VS Code, JetBrains); claude.ai and Claude Desktop on paid plans, where hooks and plugin agents do not run in plain chat; Pi through a community wrapper; Codex CLI through a separate project, per the README

What it includes

"Write a paper": all agents through to a complete draft
Runs 8 phases: configuration, research, architecture, argumentation, drafting, citations and abstract, peer review, formatting. Figures are added if there are quantitative results.
Try: “Write a paper on the impact of AI on higher education quality assurance.”
Write · full
"Guide my paper": Socratic, chapter by chapter
Produces a Chapter Plan and an INSIGHT Collection. When it is unclear whether you want plan or full, plan is the default.
Read · plan
Detailed outline and evidence map
Runs the intake, literature and structure agents only.
Write · outline-only
Patch-based revision with a deterministically applied draft
Does not re-emit the whole paper. It anchorizes the draft, writes a patch, applies it with scripts/ars_apply_revision_patch.py and produces an apply report.
Runs code · revision
Turns reviewer comments into a Revision Roadmap
Also covers rebuttals and grant panel responses. A separate variant tracks comments from a real institutional committee.
Write · revision-coach
Bilingual abstract and keywords
The two abstracts are written independently, not translated. The default pair is zh-TW and English.
Write · abstract-only
Annotated bibliography and synthesis
For a systematic review, SKILL.md points to deep-research systematic-review mode instead.
Write · lit-review
LaTeX, DOCX (via Pandoc), PDF, or a new citation style
Citation conversion works between APA 7, Chicago, MLA, IEEE and Vancouver.
Write · format-convert
Citation error report
Runs the citation compliance agent. SKILL.md requires every citation to be verified by DOI or web search.
Network · citation-check
Venue-specific AI-usage disclosure
Reports REQUIRED, ACTION_ONLY, NOT_REQUIRED or UNKNOWN for the venue, using a venue policy database or a policy-anchor path.
Write · disclosure
QA of your own rebuttal draft against reviewer comments
Needs both the comments and a draft. It marks each concern addressed, partial or missing, flags risks, and writes no new response.
Read · rebuttal-audit
Five categorical dimensions, at most two rounds
Originality, methodological rigor, evidence sufficiency, argument coherence and writing quality. Unresolved items become Acknowledged Limitations.
Read · Peer review loop
Data availability, ethics, CRediT, conflicts and funding
SKILL.md marks these five statements as an IRON RULE for every paper.
Write · Mandatory inclusions

From the official docs

Permissions

Medium risk

The main risk is your unpublished manuscript leaving your machine: through lookups of your references, or through cross-model checks if you turn them on. The plugin also runs hooks and agents that can edit files.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

Manuscript text can leave for OpenAI or Google
Cross-model verification runs only when ARS_CROSS_MODEL is set and you give consent each session. docs/DATA_FLOWS.md says it can send up to manuscript content to api.openai.com, generativelanguage.googleapis.com or a base URL you set. Leave ARS_CROSS_MODEL unset to send nothing. github.com
The plugin installs hooks, agents and scripts
hooks/hooks.json runs bash scripts on session start and before every Write, Edit, MultiEdit or Bash call. The agents can write and edit files. Read the plugin before you enable it, and check docs/CONTROL_AVAILABILITY.md for what is active in your install. github.com
Reference titles and DOIs go to four indexes
The citation verification gate sends identifiers and title queries of your references to api.semanticscholar.org, api.openalex.org, api.crossref.org and export.arxiv.org. DATA_FLOWS.md says none needs an account or key. github.com
It writes local files and caches
The verification cache lives at ~/.cache/ars/verification.db with a 90-day TTL per entry. Material Passport and run ledger files are written where you name them and hold your own words, so delete them with your project. github.com
Plain chat has fewer controls
The README says hooks and plugin agents do not run in plain chat on claude.ai or Claude Desktop, and SETUP.md says the write-scope guard is untested there. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only mode is documented.
Talks to	api.semanticscholar.org, api.openalex.org, api.crossref.org and export.arxiv.org for citation lookups. raw.githubusercontent.com for the daily plugin update check. api.openai.com, generativelanguage.googleapis.com or your own base URL, only if cross-model verification is on and you consent. Pandoc and tectonic run locally when you ask for DOCX or PDF
Sign-in and keys	The suite needs no account with the ARS project. Optional keys such as OPENAI_API_KEY, GOOGLE_AI_API_KEY and S2_API_KEY are read from your environment. docs/DATA_FLOWS.md says the suite itself contains no telemetry and no analytics endpoint.
Scopes	The plugin registers two hooks: a SessionStart announce script and a PreToolUse hook (matcher Write, Edit, MultiEdit and Bash) that runs a write-scope guard The write-scope guard is optional Python. It denies out-of-scope writes and all Bash for single-phase agents that run as separate subagents, and passes everything through if no real Python is found The plugin adds 4 agents (synthesis-agent, research-architect-agent and report-compiler-agent list Write and Edit, and screening-reviewer-agent lists Read and Grep) and 16 slash commands Citation checks send reference titles and DOIs to Semantic Scholar, OpenAlex, Crossref and arXiv, and cache results in ~/.cache/ars/verification.db A SessionStart update check fetches the public plugin.json from raw.githubusercontent.com at most once every 24 hours; ARS_UPDATE_CHECK=0 turns it off

Questions

What does the academic-paper skill do?

It guides Claude through planning, drafting, citing, reviewing and formatting an academic paper, or through one step of that, such as an abstract, a citation check or a revision after reviews.

Which mode should I start with?

SKILL.md says to start with plan if you are not sure. It guides you step by step. Use full when you already have a clear research question, and revision-coach when you have reviewer comments.

Can I use it for paid or commercial work?

The license is CC BY-NC 4.0, which allows non-commercial use only. The LICENSE file, README badge and plugin.json all say so. Read the license text before using it for paid work.

Does it write the whole paper for me?

In full mode it can draft one, but it stops for your confirmation, including the Paper Configuration Record before research starts and the outline. The README says the decisions stay yours. Our advice: check every reference yourself, even though SKILL.md requires citations to be verified.

Does it only work in English?

It follows your language and keeps academic terms in English. The bilingual abstract follows a declared language pair, with Traditional Chinese plus English as the default. Trigger words exist in English, Spanish, Traditional Chinese and Korean.

How is it different from deep-research and academic-paper-reviewer?

Per SKILL.md, deep-research makes an APA 7 research report, while academic-paper makes a journal-ready draft with five citation formats and bilingual abstracts. Reviewing a paper goes to academic-paper-reviewer, and the full research-to-paper pipeline goes to academic-pipeline.

Does it need an API key?

Claude Code needs its own key, which the README lists as a prerequisite. The citation lookups work without an account or key, per docs/DATA_FLOWS.md. Cross-model verification is optional and needs an OpenAI or Google key.

Does it work in Codex?

Not through this skill's install box. The README sends Codex CLI users to a separate project, Imbad0202/academic-research-skills-codex. This page does not cover it, so check its README.

Is it safe?

The skill folder bundles no scripts, only a LaTeX template. The plugin route adds hooks, scripts, 16 slash commands and 4 agents. Your manuscript goes through your Claude connection, and it leaves for another provider only if you turn on cross-model verification and consent. Read the safety section on this page.

Information

updated Nov 2, 2026

Repository: imbad0202/academic-research-skills
Folder: academic-paper

imbad0202/academic-research-skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 6

academic-research-skills README · github.com
docs/SETUP.md · github.com
plugin.json · github.com
academic-paper SKILL.md · github.com
academic-paper SKILL.md · github.com
references/mode_selection_guide.md · github.com